The benches that you've done make it difficult to compare because they were all gathered at different CPU frequencies. It makes it look like this memory is going to make a huge difference but most of that difference is because you're testing at totally different CPU frequencies.
For a better comparison you need to keep the CPU MHz constant. No one buys high performance memory like this so they can run it at CL3 so why not concentrate on just CL4 and CL5 timings. With a 10x multiplier CPU I'd test at:
10.0 x 400 MHz CL4, CL5
8.0 x 500 MHz CL4, CL5
7.5 x 533 MHz CL5
This way you're keeping the CPU speed constant at ~4000 MHz so any difference in performance is due to the memory and not the CPU running faster. You could then use a couple of different memory dividers, whatever works with your combo, but at least you would be able to see the real difference big DDR2 memory speeds make to overall performance.
You also need to show some stability benchmarks. There's no use having memory that can run some nice Everest bandwidth numbers at high DDR2 speeds if it's producing memory errors in MemTest86+ or crapping out in SuperPi / HyperPi 32M, OCCT or Prime95.
Not trying to be a dink. Just some suggestions to make your next memory review even better.